Paper 3

Transcription

Paper 3
Paper 3
Local site effects in Ataköy,
Istanbul, Turkey, due to a future
large earthquake in the
Marmara Sea
Sørensen, M.B., Oprsal, I., BonnefoyClaudet, S., Atakan, K., Mai, P.M., Pulido, N.
and Yalciner, C.
Accepted for publication in Geophysical
Journal International
Accepted for publication in Geophysical Journal International (March 2006)
Local site effects in Ataköy, Istanbul, Turkey, due to a future large earthquake in
the Marmara Sea
Mathilde B. Sørensen1, Ivo Oprsal2,3, Sylvette Bonnefoy-Claudet4,5, Kuvvet Atakan1, P.
Martin Mai2, Nelson Pulido6 and Caglar Yalciner7,8
1Department
2
of Earth Science, University of Bergen, Norway,
Institute of Geophysics, Swiss Federal Institute of Technology ETH, Zurich,
Switzerland,
3
Now at Charles University, Faculty of Mathematics and Physics, Department of
Geophysics, Prague, Czech Republic,
4LGIT,
5
Université Joseph Fourier, Grenoble, France,
Now at Department of Astronomy, Physics of the Earth and Meteorology, Comenius
University, Bratislava, Slovak Republic,
6
Earthquake Disaster Mitigation Research Center EDM, NIED, Kobe, Japan,
7Osmangazi
8
University, Eskisehir, Turkey,
Now at EOST-IPG, Strasbourg, France.
Accepted . Received
; in original form
.
Abbreviated title: Local site effects in Ataköy, Istanbul, Turkey.
Corresponding author: Mathilde Bøttger Sørensen, University of Bergen, Department of
Earth Science, Allegt. 41, 5007 Bergen, Norway, Tel: +47 55588755, Fax: +47
55583660, e-mail: [email protected].
1
Accepted for publication in Geophysical Journal International (March 2006)
Summary
Since the 1999 Izmit and Duzce earthquakes in northwest Turkey, many seismic hazard
studies have focused on the city of Istanbul. An important issue in this respect is local site
effects: strong amplifications are expected at a number of locations due to the local
geological conditions. In this study we estimate the local site effects in the Ataköy area
(southwestern Istanbul) by applying several techniques using synthetic data (hybrid 3D
modelling and 1D modelling) and comparing to empirical data. We apply a hybrid 3D
finite-difference (FD) method that combines a complex source and wave propagation for
a regional 1D velocity model with site effects calculated for a local 3D velocity structure.
The local velocity model is built from geological, geotechnical and geomorphological
data. The results indicate that strongest spectral amplifications (SA) in the Ataköy area
occur around 1 Hz and that amplification levels are largest for alluvial sites where SA
reaching a factor of 1.5-2 can be expected in case of a large earthquake. We also compare
our results to H/V spectral ratios calculated for microtremor data recorded at 30 sites as
well as to ambient noise synthetics simulated using a 1D approach. Because the applied
methods complement each other, they provide comprehensive and reliable information
about the local site effects in Ataköy. Added to that, our results have significant
implications for the southwestern parts of Istanbul built on similar geological formations,
for which therefore similar SA levels are expected.
Keywords: Earthquakes, Hybrid method, Strong ground motion, Istanbul, Fault model,
Finite-difference methods
2
Accepted for publication in Geophysical Journal International (March 2006)
Introduction
Istanbul, with a population exceeding 12 millions, is considered one of the worlds megacities exposed to a large earthquake hazard. The catastrophic consequences of the two
large earthquakes in Izmit and Düzce in 1999 have highlighted the need for careful
analysis of seismic hazard including local site effects, although the earthquake hazard in
this region has been a topic of considerable interest for a long time. Recent results from
several studies (e.g. Atakan et al. 2002, Erdik et al. 2003, Erdik et al. 2004, Pulido et al.
2004) emphasize the importance of earthquake preparedness and risk mitigation in the
Istanbul metropolitan area and its rapidly growing surroundings. The present study
addresses the issue of local site effects in this area.
Previous studies of local site effects, following the 1999 Izmit and Duzce earthquakes,
have focused mainly on the Avcilar district of western Istanbul (e.g. Özel et al. 2002,
Tezcan et al. 2002), and on the city of Adapazari in the east (e.g. Bakir et al. 2002,
Komazawa et al. 2002, Sancio et al. 2002, Beyen & Erdik 2004 and Ansal et al. 2004),
which experienced significant damage mainly due to site effects. In both areas, the
presence of soft sediments in basin structures has caused strong amplification of
earthquake ground motion during past earthquakes.
As for the city of Istanbul, less attention has been paid to the possible effects of local
geological variations. In a recent study, Birgören et al. (2004) found amplification levels
3
Accepted for publication in Geophysical Journal International (March 2006)
up to a factor of 7 for some geological formations at 1 and 3 Hz frequencies, based on
spectral ratios of records from a M=4.2 earthquake.
The main objective of this study is to estimate the local site effects in the Ataköy district
of the Bakirköy Municipality, which lies to the east of the Atatürk international airport in
southwestern Istanbul. In general, the area is geologically representative of the
southwestern part of Istanbul (Figure 1a), and the results therefore give an insight into the
site conditions in this larger-scale area. An important argument for focusing on this area
stems from recent results of ground motion modelling (Pulido et al. 2004) where the
highest ground motions due to a scenario earthquake in the Marmara Sea are expected in
the southwestern part of Istanbul (Figure 1b). This trend is also supported by probabilistic
seismic hazard results for this area, predicting the highest seismic hazard in the southern
parts of the city due to the close vicinity to the North Anatolian fault (Atakan et al. 2002).
In addition to that, the area is densely populated, including critical facilities such as the
Atatürk international airport and several industrial installations.
In this study, we apply three parallel approaches for assessing the local site effects in the
Ataköy region. We express the site effects in terms of amplification describing the ratio
of the ground motion at the free surface to that at bedrock level. Since our final goal is to
understand the site effects in three dimensions, we use a hybrid FD procedure to calculate
spectral amplification due to a 3D velocity-density model representing the Ataköy area.
In order to improve this model and obtain further insight into the local variations of site
4
Accepted for publication in Geophysical Journal International (March 2006)
effects, H/V spectral ratios are calculated for recorded microtremor data and compared to
H/V spectral ratios for synthetic ambient noise data based on 1D geological models.
The applied hybrid 3D FD procedure for calculating spectral amplifications (Oprsal &
Zahradnik 2002, Oprsal et al. 2002) combines source, path and site effects in two
consecutive steps. In the first step, input ground motions for bedrock conditions are
modelled for a target scenario earthquake and a regional 1D crustal structure (Pulido et
al. 2004). In the second step, a 3D FD scheme is used for calculating the spectral
amplifications within the study area for a local 3D velocity structure. The procedure is
capable of computing the full 3D wave field propagation. The resulting ground motion
and corresponding SA factors (related to a bedrock site) cover the local area through
surface receivers placed on a fine square grid. This kind of modelling and comparison
provides a comprehensive picture of the site effects, although uncertainties in the input
geological model may limit the final result.
The H/V spectral ratio method (or ‘Nakamura method’) (Nakamura 1989, Nakamura
2000) is a well-established method for fast and cost-efficient estimation of site effects.
The limitations of the methodology are studied in detail in previous review papers (e.g.
Lachet & Bard 1994, Atakan 1995, Kudo 1995, Mucciarelli 1998, Mucciarelli et al. 2003,
Atakan et al. 2004a). General consensus is that when the method is applied under careful
experimental conditions (i.e. type of the instruments used, measurement details with
regard to the sensor coupling, surface conditions, external factors such as wind, rain etc.,
and the processing techniques used for analysis), fundamental frequencies on which the
5
Accepted for publication in Geophysical Journal International (March 2006)
amplifications occur can be reliably assessed. However, the method is less reliable for the
absolute amplification factors.
Synthetic noise data based on 1D models representative of the Ataköy area provide
additional information about the effect of varying structures on the site amplification, and
thus helps in interpreting the microtremor results. The method used to model the
synthetic noise has recently been developed for more systematic investigation of the H/V
spectral ratio method (Bonnefoy-Claudet et al. 2004), however, the simulated noise also
provides an input for studies of the effect of different velocity structures on the H/V ratio.
As opposed to other analytical methods for estimating the ground response such as with
one-dimensional layered damped soil on elastic bedrock using earthquake records (e.g.
SHAKE, Schnabel et al. 1972), the method used in this study takes advantage of the noise
wavefield and therefore provides a more appropriate basis for correlating the empirical
results from the H/V spectral ratios.
Due to the differences in their uncertainties, the applied methods complement each other
well, and in combination give a reliable estimate of the local site effects in the Ataköy
area.
Geological setting and site selection
The metropolitan area of Istanbul is underlain by Paleozoic bedrock outcropping in the
northern part of the city (north of Golden Horn) with alluvial systems of Quaternary age
6
Accepted for publication in Geophysical Journal International (March 2006)
dissecting into the bedrock (Figure 1a). In the southwestern part of the city, on the other
hand, weaker geological formations are dominating such as the Bakirköy and Güngören
formations, with significant interplay of alluvial and delta systems. Taking into account
this broad perspective, our attention is focused on the southwestern part of Istanbul,
where the strongest local site effects are expected.
In general, dominating geological formations in southwestern Istanbul are the Bakirköy
and Güngören formations, which are both of upper Miocene age. The Bakirköy formation
is composed of alternating layers of limestone, marl and clay, whereas the Güngören
formation consists of green coloured plastic clay, marl and clayey siltstone. These are
also the dominating formations in the target area for this study, the Ataköy district. Figure
2 shows a detailed geological map of the Ataköy area. In addition to the Bakirköy and
Güngören formations, the Kusdili formation of Quaternary age outcrops in a limited area
and is composed of clay with molluscs, silt and mud. The overlying alluvial deposits
(Quaternary) are the result of fluvial activity and consist of unconsolidated sediments
composed of gravel, sand, silt and clay. In some parts of the area, construction material is
dumped over the alluvium, overlain by a thin layer of gravel (20-30cm) and filled with
soil on top (40-50 cm). The total thickness of these deposits is approximately 2-3 meters.
Our study focuses on the Bakirköy and Güngören formations because of their large
spatial extent, as well as on the alluvium where strong site amplifications are expected to
occur.
7
Accepted for publication in Geophysical Journal International (March 2006)
3D modelling of local site effects
The frequency-dependent ground response in the Ataköy area, due to a finite-extent
source, regional-model path effects, and detailed local structure, is calculated using the
hybrid FD procedure of Oprsal & Zahradnik (2002) and Oprsal et al. (2002). We use the
hybrid formulation to model the complete wavefield because direct FD computations for
a frequency range of engineering interest are too demanding in terms of computer
memory and time. As for the methods used for the wave propagation modelling in
complex 3D media, an FD method is considered one of the most appropriate means for
complete wave field simulation because of its simplicity, stability and relatively simple
implementation. To decrease the time and memory demands, our hybrid method
computes the wave field by FD in a local model containing complex 3D structure
embedded in a (usually simpler and smoother) regional-structure medium. Combination
of the source, path and local site effects in a two-step procedure (Figure 3), was first
introduced by Alterman & Karal (1968) for representation of a seismic source in 2D
simulation. The present method is a 3D/3D FD version, which is described in detail in
Oprsal & Zahradnik (2002), Oprsal et al. (2002) and Oprsal et al. (2005). Here we only
give a short summary.
In the first step, ground motion time series are calculated for receivers covering the
surface of a double-planed box (called excitation box (EB)) surrounding the local site of
interest. These calculations, for a seismic source placed outside the EB, are performed for
a regional crustal velocity model with an outcrop Vs of 2.0 km/s, using the methodology
8
Accepted for publication in Geophysical Journal International (March 2006)
of Pulido et al. (2004). This is based on a procedure combining a deterministic simulation
at low frequencies (0.1-1 Hz) with a semi-stochastic simulation at high frequencies (1-10
Hz). A finite-extent scenario earthquake source embedded in a flat-layered 1D velocity
structure is assumed. The source consists of a number of asperities, which are divided
into subfaults assumed to be point sources. The total ground motion at a given site is
obtained by summing the contributions from the different subfaults. For the low
frequencies, subfault contributions are calculated using discrete wave number theory
(Bouchon 1981) and summed assuming a given rupture velocity. At high frequencies, the
subfault contributions are calculated using the stochastic method of Boore (1983) and
summed using the empirical Greens function method of Irikura (1986). The radiation
pattern is changed from a theoretical double-couple radiation pattern at low frequencies
to a uniform radiation pattern at high frequencies following Pulido & Kubo (2004). The
resulting wavefield for receivers on the EB is saved on disk.
In the second step, we perform a hybrid wave field injection of the excitation computed
in the 1st step into the 2nd step 3D FD method. The 3D FD modelling now comprises
detailed local structure and occupies only a small fraction of the model size considered in
the first step. This approach benefits from the efficiency of the less demanding sourceand-path-effects methods while exploiting the wave field completeness of the FD
method. Thus final response contains the combined effects of source, path, and site
effects while the memory and time requirements are still in realistic bounds.
9
Accepted for publication in Geophysical Journal International (March 2006)
The hybrid coupling keeps the excitation boundary fully transparent in the second step.
The scattered wave field penetrates freely out of the EB and, if reflected by an
inhomogeneity, it freely propagates through the EB back into the local structure. The
same applies for possible new sources added in the second (FD) step (Oprsal &
Zahradnik 2002, Oprsal et al. 2002, Oprsal et al. 2005).
The scenario earthquake used in the first step of the calculations is a slight modification
of the worst-case scenario for the city of Istanbul defined by Pulido et al. (2004), which
assumes a combined rupture of the two segments of the North Anatolian Fault in the
Marmara Sea (total fault length 130 km) in a M=7.5 earthquake with rupture initiation in
the westernmost end. Fault asperities are located in the central part of the rupturing fault,
close to the boundary between the two fault segments, considering the seismicity in the
area. Stress drop is calculated using the results of Das & Kostrov (1986) to 50 bar
(background) and 100 bar (asperity), whereas rise time (3s) and rupture velocity (varying
randomly between 2.8-3.2 km/s) are based on the values for the 1999 Izmit earthquake.
We use the 1D regional crustal velocity model, which is used for routine earthquake
location in the Marmara Sea region (Serif Baris, personal communication, 2003). From
this scenario earthquake, the ground motions were calculated on a coarse regular grid
covering the Marmara Sea area.
As input for the second step of the computations, a densely sampled wavefield on the
excitation box is needed. This was obtained by interpolating the above described coarsegrid regional simulation result. In order to avoid aliasing effects when creating the
10
Accepted for publication in Geophysical Journal International (March 2006)
densely sampled wavefield (spacing 0.25 m) from the coarsely gridded input data
(spacing 10 km) we used a Fourier-domain resampling approach. Waveforms were first
aligned with respect to propagation direction using a cross-correlation approach, and then
spatially interpolated on to the fine grid. The time-shifts for the input waveforms were
interpolated to the fine grid as well, and then used to undo the waveform alignment.
The local 3D geological model is based on available geological, geotechnical and
geomorphological data. The data sources are microtremor measurements, standard
penetration test (SPT) data, cone penetration test (CPT) data, geological and
geomorphological maps and empirical relations for rock characteristics (Schön 1996).
The geometry of the model is determined from the geological and geomorphological
maps of the area, and a number of assumptions have been made. The surface geometry is
shown in Figure 4 together with the locations of recording sites and boreholes. The
alluvial layer in the model is maximum 5 m thick in all places except under the sea where
it reaches 10 m. The alluvium is assumed to be deposited in the depression caused by
erosion of the Bakirköy formation due to fluvial activity (i.e. there is no change of the
Bakirköy formation base under the alluvium). The Bakirköy formation is 20 m thick,
thinning northwards as it erodes at the surface. In the eastern part of the model, the
formation is slightly thicker in order to incorporate the increased elevation. The
Güngören formation is 80 m thick, thinning in places where erosion occurs. A number of
EW cross-sections of the resulting model are shown in Figure 5, and in Figure 6 is shown
elevation maps of the different layers in the model.
11
Accepted for publication in Geophysical Journal International (March 2006)
The velocity model of the area is built on the geological model and gives Vp, Vs and
density as a function of depth. Quantification of the velocities for the 3D grid is based on
the formulae given in Table 1. For the alluvium, a low surface velocity and a low depth
gradient is used. The chosen S-wave velocity is consistent with a NEHRP class E soil
(soft soil) at the surface compacting to a class D soil (stiff soil) at 10 m depth. Due to
heavy exploitation of water through boreholes in the Ataköy area, the present ground
water level is at 100 m below ground level (ITU Gelistirme Vakfi, 2000). In general in
the larger Ataköy-Bakirköy region, ground water level varies between 60-160 m. This
implies that we can assume dry soil conditions for the alluvial layer, and for simplicity
we assume a Vp/Vs ratio of 1.732. For the Bakirköy and Güngören formations, low
surface velocities are chosen. For the upper 5 m there is a large velocity gradient for these
formations, whereas the gradient is much lower for depths larger than 5 m. The low
surface velocities and high gradients for the uppermost meters are included to take into
account the strong weathering taking place at the surface. For the bedrock, the velocity
structure is chosen in order to be consistent with the velocity model used in step 1 of the
modelling (Pulido et al. 2004) at depths larger than 150 m. In the formulae of Table 1,
input depths are relative to the free surface, implying that the velocity i.e. at the top of the
Bakirköy formation is dependent upon the depth of the overlaying alluvium. This is to
take into account the increased pressure arising from increased overburden.
The dimensions of the site of interest, given by the geological model, are 3360 m and
2219 m in the EW and NS directions, respectively, and approximately 180 m in depth.
The S-wave velocities Vs are between 260 m/s and 2100 m/s, and maximum P-wave
12
Accepted for publication in Geophysical Journal International (March 2006)
velocity Vp = 3600 m/s. These high variations in material parameters impose strong
demands on the computational part of the FD problem. The influence of the water layer is
neglected, which is supposed not to be playing a major role in the resulting wave field
since it is only present in the very southernmost part of the model. Therefore FD
computation of the site effects was performed on a vertically irregular grid with grid steps
being 1.5 m in the vicinity of the free surface and 21 m in deeper parts of the model. The
horizontal spacing of the grid is 2.5 m and remains unchanged through the whole
visualized part of the model because of the spatial distribution of low-velocity riverbeds.
However, to minimize the spurious reflections, the model was extended to each side and
depth by 70 grid points where the grid step increases towards the edges of the
computational model. At this part, where the known model is extended, we gradually
decrease the depth of the interfaces and increase the grid step. To decrease spurious
reflections, we apply tapers (Cerjan et al. 1985) at the strip of 50 grid points breadth
around the edges, and non-reflecting boundaries at the edges of the computational model
(Emerman & Stephen 1983). The PML technique was taken into consideration in place of
non-reflecting boundaries, however, these are too computationally demanding in our
formulation. The frequency band is 0-8 Hz and the time step determined from the
minimum ratio of DX(x,y,z)/(1.6*Vp(x,y,z)) is DT= 4.0 x 10-4 s. The computational
model dimensions, further extended by the taper zones, are then 4600 m and 3500 m in
the EW and NS directions, respectively, and approximately 1500 m in depth. The total
number of the grid points is 1.5 x 108, occupying approximately 5.9 GB of core memory,
which is on the edge of reasonable time demand for a 2-processor PC. Because of very
modest topography, the model in the computations has a flattened free surface. The
13
Accepted for publication in Geophysical Journal International (March 2006)
flattening shifts the vertical structural profile under each point on the free surface so that
such a profile remains unchanged.
To provide a more complete picture of the site ground motions on potential buildings, we
give the results in the spectral amplification. The spectral amplification factors for pseudo
acceleration response (PSA) depending on the frequency are shown in Figure 7a. Shown
PSA factors cover the whole area and they are computed as ratio of 5%-damping spectral
responses computed for the 3D and 1D (bedrock) models, respectively. The maximum
PSA amplifications for the alluvial sites are well above 1.5, especially at frequencies
above 3 Hz. Lower amplification levels, less than 1.5, are uniformly distributed over the
entire area at all frequencies (Figure 7a). A more pronounced amplification can be
expected for frequencies between 10-20 Hz due to the shallow low-velocity deposits.
To compare the 3D FD modelling with a possibly more sophisticated 1D linear method,
we performed a series of approximately 3 x 105 1D-structure-response computations
(Mueller 1985) for points regularly distributed on the free surface. The 1D structure for
each point was exactly the same as the vertical profile under such a point in the 3D
model, hence pseudo 3D modelling. The code for 1D response computation in layered
media was adopted after Bartak & Zahradnik (1991). The PSA factors are shown in
Figure 7b. The maximum amplification for the alluvial systems is approximately 1.5. The
significant amplification at these sites is well visible for all frequency bands, which is the
main difference between the 1D and 3D response computation.
14
Accepted for publication in Geophysical Journal International (March 2006)
H/V spectral ratios of microtremor data
Recorded microtremor data give the opportunity of assessing the fundamental frequency
at a given site based on the H/V (horizontal to vertical component of the recorded signal)
spectral ratio technique (also known as ”Nakamura method”) (Nakamura 1989,
Nakamura 2000, Lermo & Chávez-García 1993). This method is extensively used in
microtremor studies throughout the world. The method is discussed in detail in previous
studies and the reader is referred to the literature on this topic. Some recent discussions
are given in Atakan et al. (2004a, 2004b), Guillier et al. (in prep.) and Chatelain et al. (in
prep.).
The H/V spectral ratio method is based on the assumption that amplification of
microtremor ground motion, mainly consisting of Rayleigh wave energy, due to the
presence of a soft surface soil layer, only occurs for the horizontal component of ground
motion. Under this assumption, the vertical component of ground motion can be used to
remove source and path effects from the signal and isolate the effect of the site (Lermo &
Chávez-García 1993). The microtremor recordings are transformed into the frequency
domain and the horizontal spectrum is divided by the vertical spectrum. The resulting
spectral ratio gives frequency dependent amplification for the site.
In spite of the known limitations of the method, various sets of experimental data (e.g.
Field & Jacob 1993, Duval et al. 1994, Duval et al. 1995, Field et al. 1995, Larchet et al.
1996, Gitterman et al. 1996, Fäh et al. 1997, Lebrun et al. 1997, Riepl et al. 1998)
15
Accepted for publication in Geophysical Journal International (March 2006)
confirm that the spectral ratios are much more stable than the raw noise spectra and
exhibit a clear peak at soft soil sites, which is well correlated with the fundamental
resonance frequency. These observations are supported by several theoretical
investigations (Field & Jacob 1993, Lachet & Bard 1994, Lermo & Chavez-Garcia 1994,
Bonnefoy-Claudet et al. 2004), showing that synthetics obtained with randomly
distributed, near surface sources lead to H/V ratios sharply peaked around the
fundamental S-wave frequency, whenever the surface layers exhibit a sharp impedance
contrast with the underlying stiffer formations.
During a field campaign in the Ataköy area, microtremor data were collected at 30 sites
(Figure 2). The sites were located mainly on the alluvium and the Bakirköy formation.
Site selection was based on avoiding too much man-made noise and at the same time
obtaining good coupling of the sensor to the ground. At each site, a minimum of 3 x 10
minutes of seismic noise was collected continuously using the GBV-316 (GeoSIG)
portable digital seismographs. Each seismograph contains a 16-bit digitizer and a 4.5 Hz
3-component built-in sensor. The technical specifications of the instruments used were
tested extensively through previous studies (Atakan et al. 2004a) and have a resolution
down to 0.5 Hz (also down to 0.3 Hz under specific conditions). Communication was
done using the Seislog data acquisition software developed for Pocket PC (Ojeda et al.
2004).
Data were processed using the recently available software J-SESAME (Atakan et al.
2004b), developed for calculation of H/V spectral ratios. An automatic window selection
16
Accepted for publication in Geophysical Journal International (March 2006)
module is included, which filters out noisy time windows by applying an anti-trigger
algorithm based on the STA/LTA ratio (short term average divided by the long term
average of the signal amplitudes). The data are organized according to the recording sites
and average H/V spectral ratios are computed using standard processing techniques
(Atakan et al. 2004b).
The H/V spectral ratios calculated for sites on the Alluvium and Bakirköy formation are
presented in Figures 8 and 9, respectively. For the alluvial sites (Figure 8), a strong peak
is observed around 1 Hz. An additional, more diffuse peak is indicated around 3-6 Hz.
For the Bakirköy formation (Figure 9), there is again a clear peak around 1 Hz whereas
no peaks are observed for higher frequencies. This indicates that the 3-6 Hz peak
observed for the alluvium is an effect of the alluvial layer, whereas the 1 Hz peak is
probably caused by deeper lying formations.
In a previous study by Eyidogan et al. (2000), microtremor recordings were collected at a
few sites in the Ataköy area (Figure 4). H/V ratios for these data, recorded mainly on
alluvial deposits, are in agreement with the results obtained in the present study.
1D modelling of ambient noise
In order to check the H/V spectral ratio results for the recorded microtremors, ambient
noise was simulated using 1D models representative of sites in Ataköy. H/V spectral
ratios for the simulated noise were then calculated for comparison with the H/V spectral
17
Accepted for publication in Geophysical Journal International (March 2006)
ratios of the recorded microtremors. The noise simulations were performed as described
by Bonnefoy-Claudet et al. (2004), simulating noise originated by human activity for
sites with heterogeneous subsurface structure. In this study, all sites are considered as 1D
structures, and the Green’s functions for the medium are calculated using the method of
Hisada (1994). H/V spectral ratios for the synthetic noise were calculated using the JSESAME software as described for the recorded microtremors.
Ambient noise was simulated for three sites representative of an alluvial site, a site on the
Bakirköy formation and a site on the Güngören formation. The 1D structures of the sites
are taken from the 3D model used for 3D FD simulations. The velocity profiles for the
1D sites were simplified so that each layer has a constant velocity to obtain reasonable
computation times. The composition of these sites is illustrated in Figure 10 and the
velocities are given in Table 2.
For each site, noise sources were modelled as point forces with delta-like source time
functions located at fixed depths (4 and 8 m) and distributed randomly in space, direction,
amplitude and time. Convolving Green’s functions calculated by the method of Hisada
(1994) gave synthetic noise representative of the three sites.
The resulting H/V spectral ratios are shown in Figure 11. The significant peak around 1
Hz observed for the recorded microtremors is not clearly present for the model results.
Instead, a much more diffuse peak is observed around 1-3 Hz with smaller amplitude
than for the recorded microtremors. In addition, a significant peak is present for the
18
Accepted for publication in Geophysical Journal International (March 2006)
alluvial site (site 1) at higher frequencies. This peak is probably related to the 3-6 Hz
peak observed for the recorded microtremors. There is generally a shift of the modelled
peaks towards higher frequencies. This indicates uncertainties in the layer thicknesses or
impedance contrasts in the models. Amplification levels are in agreement with those for
the recorded microtremors, except for the 2-3 Hz peak, which is probably smaller due to
simplifications in the modelling.
Comparison and discussion
Individual results obtained from the empirical data and their comparison to 1D and 3D
synthetic modelling show the following features:
•
3D synthetic results give an insight to the complexity of the site response,
especially for higher frequencies where lateral variations become more visible. In
this respect the response of the alluvium is clearly visible at frequencies higher
than 2 Hz.
•
Regarding the frequency content of the site effects, the clear peaks observed in the
microtremor data around 1 Hz are comparable to 3D synthetic results. However,
some differences are observed due to uncertainties in the velocity model, which is
based on very limited available data. The 1Hz peak is in good agreement with
what might be expected from the soil thickness using the relation fpeak=vs,ave/4H
where vs,ave is average S-wave velocity and H is thickness of the soil layer.
19
Accepted for publication in Geophysical Journal International (March 2006)
•
The different methodologies predict different values for the amplification factors.
The noise-based methodologies (recorded microtremors and 1D modelling) have
significantly higher amplification levels when compared to the synthetic 3D data.
•
Our results from the 3D modelling, based on a simulated strong ground motion
(scenario earthquake of magnitude 7.5 in the Marmara Sea) predict maximum
amplification factors in the range 1.5-2. This is significantly lower than our results
from microtremor data and previous results on weak motion data (Birgören et al.
2004).
•
One explanation for the lower amplification factors for the synthetic 3D data
relative to the microtremor measurements is that the real structure is definitely
more complicated than our models, resulting in a poorer wavefield in terms of
scattering for the modelling.
•
The discrepancies between the site effect estimates of the various methodologies
reflect the lack of good geotechnical data in the area. With the availability of Swave velocity profiles, the input models can be improved significantly, which is
expected to lead to more consistent results. In this respect, our model could be
used as a starting model for refinement in future studies.
•
Based on the above, it is recommended that site effects in the Ataköy area should
be taken into account in any future application towards earthquake risk mitigation.
Our results indicate that for a target earthquake of magnitude 7.5 in the Marmara
Sea, a minimum amplification factor of 1.5 within the frequency band of 0.5-1.5
Hz is expected.
20
Accepted for publication in Geophysical Journal International (March 2006)
The study area is covered mainly by the Bakirköy formation, which consists of
alternating layers of limestone, marl and clay of Upper Miocene age. This formation is
quite fragmented and altered at the surface. The underlying Güngören formation has
similar characteristics with respect to the lithology. The alluvium, on the other hand,
represents the most critical unit in terms of site amplifications and is limited to the fluvial
depositional centres. The proximity to the coast of the area influences the lithological
characteristics of both the sediments and sedimentary rocks. The gentle topography of the
area, with shallow synclines and anticlines plunging towards the Marmara Sea in the
south, represents an environment, which is significantly different from classical alluvial
valleys or closed sedimentary basins. In this respect, the expected site effects also differ
significantly. Modelling such an environment in 3D presents several challenges with
respect to the seismic velocities and their lateral variations. The 3D model outlined in this
paper therefore introduces significant constraints in seismic wave propagation. In many
respects the model resembles to a simplified 2D approximation due to the “open-ended
architecture” of the system where the lateral extent of the sedimentary units are
continuous over large distances. The 3D FD computation of PSA amplification shows a
remarkable recemblance to the 1D pseudo 3D method.
Despite the uncertainties in the input 3D velocity model for the FD simulations, results
are in reasonable agreement with the H/V spectral ratio results based on recorded
microtremor data. The fact that independent studies based on completely different data
and methodologies give results in agreement supports the validity of conclusions drawn
from these results.
21
Accepted for publication in Geophysical Journal International (March 2006)
Implications and conclusions
In this study, local site effects in the Ataköy district of western Istanbul have been studied
using three different approaches. A hybrid FD method was applied to calculate
amplification levels on a fine grid covering the entire Ataköy area based on a local
geological model. This modelling indicates that amplification levels are highest for the
alluvial sites, where amplification up to a factor of two is predicted. H/V spectral ratios of
recorded microtremors were determined, and revealed a dominating peak of amplification
around 1 Hz for the whole area. For the alluvial sites a more diffuse secondary peak was
observed at 3-6 Hz frequencies. These ratios were compared to H/V ratios calculated for
synthetic noise at three sites representative of Ataköy. The synthetic data generally
indicated peaks at higher frequencies due to uncertainties related to the highly simplified
input models. This underlines the importance of good geotechnical data in site effect
assessment.
Based on the above discussions it is clear that site effects in the Ataköy area will have
significant consequences in case of future large earthquakes in the Marmara Sea.
However, other factors such as construction practices, density of building stock and
proximity to alluvial sediments will play an important role, especially when taking into
account the frequency variations of the site effects. Our results are naturally valid only
for the Ataköy area, however the similarity of the geological formations in the
neighbouring Bakirköy and Zeytinburnu districts may give an insight to possible
consequences in these highly populated areas in Istanbul. The similarity of the studied
22
Accepted for publication in Geophysical Journal International (March 2006)
region to the surrounding areas makes it possible to use the present results in a broader
context, concerning the importance of local site effects in southwestern Istanbul.
Acknowledgements
The present work was carried out as part of the EC project RELIEF (EVG1-CT-200200069). We are grateful to Serdar Akyuz and other staff and students at Istanbul
Technical University (ITU) for their logistic help and housing during the field campaign.
ITU kindly provided the geological, geotechnical and geomorphological data used in this
study. This is contribution # 1386 of the Institute of Geophysics, ETH Zurich.
23
Accepted for publication in Geophysical Journal International (March 2006)
References
Alterman Z. & Karal F.C., 1968. Propagation of elastic waves in layered media by finitedifference methods. Bull. Seismol. Soc. Am., 58, 367-398.
Ansal, A., Laue, J., Buchheister, J., Erdik, M., Springman, S.M., Studer, J. & Koksal, D.,
2004. Site characterization and site amplification for a seismic microzonation study in
Turkey, Proceedings of the 11th Intl. Conf. on Soil Dyn. & Earthquake Engng. (11th
ICSDEE) & the 3rd Intl. Conf. on Earthquake Geotech. Engng. (3rd ICEGE), January 79, Berkeley, CA, pp. 53-60.
Atakan, K., 1995. A review of the type of data and the techniques used in empirical
estimation of local site response, Proceedings of the "Fifth International Conference on
Seismic Zonation", October 17-19, 1995, Nice, France, Ouest Éditions, Presses
Académiques, Vol. II, 1451-1460.
Atakan, K., Ojeda, A., Meghraoui, M., Barka, A.A., Erdik, M. & Bodare, A., 2002.
Seismic hazard in Istanbul following the 17 August 1999 Izmit and 12 November 1999
Düzce earthquakes, Bull. Seismol. Soc. Am, 92, 466-482.
Atakan, K., Duval, A.-M., Theodulidis, N., Guillier, B., Chatelain, J.-L., Bard, P.-Y. & the
SESAME-team, 2004a. The H/V spectral ratio technique: Experimental conditions, data
24
Accepted for publication in Geophysical Journal International (March 2006)
processing and empirical reliability assessment, Proceedings of the 13th World Conference
on Earthquake Engineering, Vancouver, Canada, 1-6 August, 2004, paper no. 2268.
Atakan, K., Bard, P-Y., Kind, F., Moreno, B., Roquette, P. and Tento, A. & the SESAMEteam, 2004b. J-SESAME: a standardized software solution for the H/V spectral ratio
technique, Proceedings of the 13th World Conference on Earthquake Engineering,
Vancouver, Canada, 1-6 August, 2004, paper no. 2270.
Bakir, B.S., Sucuoglu, H. & Yilmaz, T., 2002. An overview of local site effects and the
associated building damage in Adapazari during the 17 August 1999 Izmit earthquake, Bull.
Seismol. Soc. Am, 92, 509-526.
Bartak, V. & Zahradnik, J., 1991. PROGRAM SITEF, Seismic response analysis of
horizontally layered sites, Charles University, Prague.
Beyen, K. & Erdik, M., 2004. Two-dimentional nonlinear site response analysis of
Adapazari plain and predictions inferred from aftershocks of the Kocaeli earthquake of
17 August 1999, Soil Dynamics and Earthquake Engineering, 24, 261-279.
Birgören, G., Özel, O., Fahjan, Y. & Erdik, M., 2004. Determination of site effects in
Istanbul area using a small earthquake record of the dense strong motion network, XXIX
General Assembly of the European Seismological Commision, Potsdam, Germany, poster
no. 249.
25
Accepted for publication in Geophysical Journal International (March 2006)
Bonnefoy-Claudet, S., Cornou, C., Kristek, J., Ohrnberger, M, Wathelet, M., Bard, P.-Y.,
Moczo, P., Faeh, D. & Cotton, F., 2004. Simulation of seismic ambient noise: I. Results
of H/V and array techniques on canonical models, Proceedings of the 13th World
Conference on Earthquake Engineering, Vancouver, Canada, 1-6 August, 2004, paper no.
1120.
Boore, D. M., 1983. Stochastic simulation of high frequency ground motions based on
seismological models of the radiated spectra, Bull. Seism. Soc. Am., 73, 1865-1894.
Bouchon, M., 1981. A simple method to calculate Green’s functions for elastic layered
media, Bull. Seism. Soc. Am., 71, 959-971.
Cerjan C., Kosloff R. and Reshef M., 1985. A nonreflecting boundary condition for
discrete acoustic and elastic wave equations, Geophysics, 50, 705-708.
Chatelain, J-L., Guillier, B., Duval, A-M., Cultrera, G., Cara, F., Teves-Costa, P., Atakan,
K., & Bard, P-Y., in prep. Evaluation of the influence of experimental conditions on H/V
results from ambient noise recordings. To be submitted to Bulletin of Earthquake
Engineering.
26
Accepted for publication in Geophysical Journal International (March 2006)
Duval, A.-M., Bard, P.-Y., Méneroud, J.-P. & Vidal, S., 1994. Usefulness of microtremor
measurements for site effect studies, proceedings of the 10th European Conference on
Earthquake Engineering, Vienna, Austria.
Duval, A.-M., Bard, P.-Y., Méneroud, J.-P. & Vidal, S., 1995. Mapping of site effects
with microtremors, proceedings of the Fifth International Conference on Seismic
Zonation, Nice, October 1995.
Emerman S.H. & Stephen R.A., 1983. Comment on Absorbing Boundary Conditions for
Acoustic and Elastic wave Equations, by R. Clayton and B. Enquist, Bull. seismol. Soc.
Am., 73, 661-665.
Erdik, M., Aydinoglu, N., Fahjan, Y., Sesetyan, K., Demircioglu, M., Siyahi, B.,
Durukal, E., Ozbey, C., Biro, Y., Akman, H. & Yuzugullu, O., 2003. Earthquake risk
assessment for Istanbul metropolitan area, Earthquake Engineering and Engineering
Vibration, Vol. 2, No. 1, 1-25.
Erdik, M., Demircioglu, M., Sesetyan, K., Durukal, E. & Siyahi, B., 2004. Earthquake
hazard in Marmar region, Turkey, proceedings of the 11th SDEE/ 3rd ICEGE, California,
USA.
27
Accepted for publication in Geophysical Journal International (March 2006)
Eyidogan, H., Ecevitoglu, B. & Caglar, I., 2000 (in Turkish). Bakirköy Ilcesi 1. etap
(Bakirköy-Ataköy) yerlesim alanlarinin jeolojik yapi ve depremsellik etüdü projesi,
Internal report, part 6, TC. Bakirköy Belediyesi – ITÜ Gelistirme Vakfi. 55 p.
Fäh, D., Rüttener, E., Noack, T. & Kruspan, P., 1997. Microzonation of the city of Basel,
Journal of Seismology, 1, 87-102.
Field, E.H. & Jacob, K., 1993. The theoretical response of sedimentary layers to ambient
seismic noise, Geophysical research letters, 20(24), 2925-2928, 10.1029/93GL03054.
Field, E.H., Clement, A.C., Jacob, K.H., Ahronian, V., Hough, S.E., Friberg, P.A.,
Babaian, T.O., Karapetian, S.S., Hovanessian, S.M. & Abramian, H.A., 1995. Earthquake
site-response study in Giumri (formerly Leninakan), Armenia, using ambient noise
observations, Bull. seismol. Soc. Am., 85, 349-353.
Gitterman, Y., Zaslavsky, Y., Shapira, A. & Shtivelman, V., 1996. Emperical site
response evaluations: case studies in Israel, Soil Dynamics and Earthquake Engineering,
15, 447-463.
Guillier, B., Atakan, K., Chatelain, J-L., Havskov, J., Duval, A-M., Ohrnberger, M.,
Cara, F., & Zacharapoulous, S., in prep. Influence of instruments on the H/V spectral
ratios of ambient vibrations. To be submitted to Bulletin of Earthquake Engineering.
28
Accepted for publication in Geophysical Journal International (March 2006)
Hisada, Y., 1994. An efficient method for computing Green’s functions for a layered
half-space with sources and receivers at close depths, Bull. seismol. Soc. Am., 84, 145672.
ITU Gelistirme Vakfi, 2000. Bakirköy Ilcesi 1. etap (Bakirköy-Ataköy) yerlesim
alanlarinin jeolojik yapi ve depremsellik etüdü projesi (in Turkish), Internal report,
chapter 9, TC. Bakirköy Belediyesi – ITÜ Gelistirme Vakfi, ITU Maden Fakultesi.
Irikura, K., 1986. Prediction of strong acceleration motion using empirical Green’s
function, Proceedings of the 7th Japan. Earthq. Eng. Symp., 151-156
Komazawa, M., Morikawa, H., Nakamura, K., Akamatsu, J., Nishimura, K., Sawada, S.,
Erken, A. & Onalp, A., 2002. Bedrock structure in Adapazari, Turkey – a possible cause
of severe damage by the 1999 Kociaeli earthquake, Soil Dynamics and Earthquake
Engineering, 22, 829-836.
Kudo, K., 1995. Practical estimates of site response, State-of-the-art report, proceedings
of the Fifth International Conference on Seismic Zonation, Nice, October 1995.
Lachet, C., & P.-Y. Bard, 1994. Numerical and theoretical investigations on the
possibilities and limitations of the "Nakamura's" technique, Journal of Physics of the
Earth, 42-4, 377-397.
29
Accepted for publication in Geophysical Journal International (March 2006)
Lachet, C., Hatzfeld, D., Bard, P.-Y., Theodulidis, N., Papaioannou, C. & Savvaidis, A.,
1996. Site effects and microzonation in the city of Thessaloniki (Greece): comparison of
different approaches, Bull. seismol. Soc. Am., 86, 1692-1703.
Lebrun, B., Bard, P.-Y. & Hatzfeld, D., 1997. Evidence for large, low frequency site
effects in large alpine valleys: the example of Grenoble, France, 29th general assembly of
IASPEI, Workshop W16, Thessaloniki, Greece.
Lermo, J. & Chávez-García, F.J., 1993. Site effect evaluation using spectral ratios with
only one station. Bull. seismol. Soc. Am., 83, 1574-1594.
Lermo, J. & Chávez-García, F.J., 1994. Are microtremors useful in site response
evaluation?, Bull. seismol. Soc. Am., 84, 1350-1364.
Mucciarelli, M., 1998. Reliability and applicability of Nakamura’s technique using
microtremors: An experimental approach, J. Earthquake Eng, 2, 1–14.
Mucciarelli, M., Gallipoli, M.R. & Arcieri, M., 2003. The stability of Horizontal-toVertical Spectral Ratio of triggered noise and earthquake recordings, Bull. Seismol. Soc.
Am., 93, 1407-1412.
Mueller, G., 1985. The reflectivity method: a tutorial. J. Geophys., 58, 153-174.
30
Accepted for publication in Geophysical Journal International (March 2006)
Nakamura, Y., 1989. A method for dynamic characteristics estimation of subsurface
using microtremors on the ground surface. Q. Rep. Railway Tech. Res. Inst., 30, 1.
Nakamura, Y., 2000. Clear identification of fundamental idea of Nakamura’s technique
and its applications, proceedings of the 12th WCEE, New Zealand, paper no. 2656.
Ojeda, J. Å., Natvik, Ø., Berland, O. & Fjeldstad, F., 2004. SEISLOG CE (SEISLOG for
Pocket PC), 29th general assembly of the European Seismological Commission,
Potsdam, Germany 12-17 September 2004, poster no. 77.
Oprsal, I., Brokesova, J., Faeh, D., & Giardini, D., 2002. 3D Hybrid Ray-FD and DWNFD Seismic Modeling For Simple Models Containing Complex Local Structures, Stud.
Geophys. Geod., 46, 711-730.
Oprsal, I. & Zahradnik, J., 2002. Three-dimensional finite difference method and hybrid
modeling
of
earthquake
ground
motion,
J.
Geophys.
Res.,
107(B8),
DOI
10.1029/2000JB000082.
Oprsal, I., Faeh, D., Mai, M. & Giardini, D., 2005. Deterministic earthquake scenario for
the Basel area - Simulating strong motion and site effects for Basel, Switzerland, J.
Geophys. Res., 110, doi:10.1029/2004JB003188.
31
Accepted for publication in Geophysical Journal International (March 2006)
Özel, O., Cranswick, E., Meremonte, M., Erdik, M. & Safak, E., 2002. Site effects in
Avcilar, west of Istanbul, Turkey, from strong- and weak.motion data, Bull. Seismol. Soc.
Am., 92, 499-508.
Pulido, N. & Kubo, T., 2004. Near-fault strong motion complexity of the 2000 Tottori
earthquake (Japan) from a broadband source asperity model, Tectonophysics, 390, 177192.
Pulido, N., Ojeda, A., Atakan, K. & Kubo, T., 2004. Strong ground motion estimation in
the Marmara Sea region (Turkey) based on a scenario earthquake. Tectonophysics, 391,
357-374.
Riepl, J., Bard, P.-Y., Hatzfeld, D., Papaioannou, C. & Nechtschein, S., 1998. Detailed
evaluation of site response estimation methods across and along the sedimentary valley of
Volvi (EURO-SEISTEST), Bull. Seismol. Soc. Am., 88, 488-502.
Sancio, R.B., Bray, J.D., Steward, J.P., Youd, T.L., Durgunoglu, H.T., Önalp, A., Seed,
R.B., Christensen, C., Baturay, M.B. & Karasayilar, T., 2002. Correlation between
ground failure and soil conditions in Adapazari, Turkey, Soil Dynamics and Earthquake
Engineering, 22, 1093-1102.
32
Accepted for publication in Geophysical Journal International (March 2006)
Schnabel, P.B., Lysmer, J. & Seed, H.B., 1972. SHAKE: a computer program for
earthquake response analysis of horizontally layered sites, Report EERC 72-12,
Earthquake Engineering Research Center, University of California, Berkeley.
Schön, J. H., 1996. Physical properties of rocks – Fundamentals and Principles of
petrophysics, Pergamon, 1996.
Tezcan, S.S., Kaya, E., Bal, I.E. & Özdemir, Z., 2002. Seismic amplification at Avcilar,
Istanbul, Engineering Structures, 24, 661-667.
33
Accepted for publication in Geophysical Journal International (March 2006)
Figure legends
Figure 1. a) Geological map of Istanbul. The black square indicates the location of
Ataköy. Redrawn from Oktay & Eren (Istanbul Metropolitan Municipality, Site Survey
and Earthquake Department). b) Peak ground velocities (PGV) predicted by Pulido et al.
(2004) for a M=7.5 scenario earthquake in the Marmara Sea. The star indicates the
rupture initiation point, the thick white lines the extent of the rupturing fault and the blue
lines the asperity locations (modified from Pulido et al., 2004).
Figure 2. Geological map of Ataköy and Bakirköy districts, western Istanbul. The study
area is outlined by the black box. Colours represent different geological formations.
Numbered dots represent the microtremor recording sites used for calculating H/V
spectral ratios. The dimensions of the study area are approximately 2.2 x 3.3 km. The
map is provided by Istanbul Technical University.
Figure 3. Schematic illustration of the hybrid procedure used for modelling spectral
amplification for a dense grid in Ataköy. a) In the first step, ground motions are
calculated on the excitation box using a regional 1D velocity model. Dark grey fields
within the fault plane are asperities. b) In the second step, surface ground motions are
calculated using a 3D finite difference scheme and a local velocity model. Figure 3b
corresponds to the stippled box in 3a.
34
Accepted for publication in Geophysical Journal International (March 2006)
Figure 4. Surface geometry of geological model of Ataköy, used in 3D FD modelling.
The black square in the uppermost map shows the extent and location of the lowermost
map. Symbols indicate borehole locations for SPT and CPT data and microtremor
recording sites used by Istanbul Technical University (ITU, Eyidogan et al. 2000) and
University of Bergen (UiB, this study). The four horizontal lines show the locations of
the profiles in Figure 5.
Figure 5. EW cross-sections of the geological model of the Ataköy area, used in the 3D
FD modelling. The locations of the profiles are shown in Figure 4. The lowermost plot is
in the southernmost part of the model, going northwards. Note the thinning of the
Bakirköy formation towards north in most of the model due to erosion.
Figure 6. Elevation plots of top of the layers of the geological model of Ataköy, used in
the 3D FD modelling.
Figure 7. a) Spectral amplification (pseudo-acceleration response PSA, damping 5%)
with respect to a bedrock site for 3D FD modelling. The results are shown for a set of
frequency bands; the left and right sides of the panel correspond to the maximum and
mean PSA amplification. The amplified response of the southern part and of the two
alluvial systems is apparent. b) Spectral amplification (pseudo-acceleration response
PSA, damping 5%) with respect to a bedrock site for the pseudo 3D (1D) modelling. The
results are shown for a set of frequency bands; the left and right sides of the panel
correspond to the maximum and mean PSA amplification. The amplified response of the
35
Accepted for publication in Geophysical Journal International (March 2006)
southern part and of the two alluvial systems is apparent. The geographical extent of the
results shown corresponds to the area shown in Figure 6. The central parts of the alluvial
systems are marked as white lines.
Figure 8. Average H/V spectral ratios for sites located on alluvium.
Figure 9. Average H/V spectral ratios for sites located at Bakirköy formation.
Figure 10. 1D models used for simulating ambient noise. a) Alluvial site: 5 m alluvial
layer, 8 m Bakirköy formation, 80 m Güngören formation and bedrock (half space). b)
Bakirköy formation site: 11 m Bakirköy formation, 80 m Güngören formation and
bedrock (half space). c) Güngören formation site: 77 m Güngören formation and bedrock
(half space). Velocities of the different layers are given in Table 4.2.
Figure 11. H/V spectral ratios calculated for simulated ambient noise at 1D sites. a)
Alluvial site, b) Bakirköy formation site, c) Güngören formation site.
36
Accepted for publication in Geophysical Journal International (March 2006)
Tables
Formation
Alluvium
Bakirköy
Güngören
Bedrock
Vs (m/s)
150 + 5 ⋅ Z’
260 + 96 ⋅ Z’ for Z’ < 5 m
685 + 11 ⋅ Z’ for Z’ > 5 m
200 + 60 ⋅ Z’ for Z’ < 5 m
445 + 11 ⋅ Z’ for Z’ > 5 m
450 + 11 ⋅ Z’
Vp/Vs
1.732
1.8
ρ (⋅1000 kg/m3)
1.7 - 1.224 ⋅ exp(-0.846 ⋅ Z’)
2.2 - 1.224 ⋅ exp(-0.846 ⋅ Z’)
1.8
2.0 - 1.224 ⋅ exp(-0.846 ⋅ Z’)
1.8
2.3 - 1.224 ⋅ exp(-0.846 ⋅ Z’)
Table 1. Formulae used for quantification of the velocities for the 3D model used for 3D
FD modelling. Vs is S-wave velocity, Vp is P-wave velocity and ρ is density. Z’ = Z –
Zfree surface is the depth relative to the free surface.
37
Accepted for publication in Geophysical Journal International (March 2006)
Thickness (m) Vs (m/s)
Alluvial site
5
163
8
784
80
1028
Half space
1473
Bakirköy formation site
11
533
80
1006
Half space
1451
Güngören formation site
77
748
Half space
1297
Vp (m/s) Density (⋅1000 kg/m3)
Formation
260
1406
1850
2799
1.169
2.691
2.500
2.700
Alluvium
Bakirköy
Güngören
Bedrock
960
1811
2757
2.078
2.500
2.700
Bakirköy
Güngören
Bedrock
1346
2464
1.878
2.700
Güngören
Bedrock
Table 2. Velocity model used for 1D modelling for alluvium, Bakirköy formation and
Güngören formation sites. The sites are taken from the model used for 3D FD modelling,
and average velocities are reported for each layer.
38
Accepted for publication in Geophysical Journal International (March 2006)
a)
b)
Figure 1
39
Accepted for publication in Geophysical Journal International (March 2006)
Figure 2.
40
Accepted for publication in Geophysical Journal International (March 2006)
Figure 3.
41
Accepted for publication in Geophysical Journal International (March 2006)
Figure 4.
42
Accepted for publication in Geophysical Journal International (March 2006)
Figure 5.
43
Accepted for publication in Geophysical Journal International (March 2006)
Figure 6
44
Accepted for publication in Geophysical Journal International (March 2006)
Figure 7
45
Accepted for publication in Geophysical Journal International (March 2006)
Figure 8.
46
Accepted for publication in Geophysical Journal International (March 2006)
Figure 9.
47
Accepted for publication in Geophysical Journal International (March 2006)
a)
b)
c)
Figure 10.
48
Accepted for publication in Geophysical Journal International (March 2006)
Figure 11.
49